Hi all,

I currently maintain a GitHub project designed to get ArangoDB up and 
running in Kubernetes.  I'm working on using StatefulSets + Persistent 
Volume Claims, so that as nodes crash and come back, they still have all of 
their same data.  I mostly have it all working, but I'm hoping some of you 
who might be familiar with how it works on the DCOS/Mesos side might have 
some answers for me.  Specifically, I have two problems I still need to 
solve:

1.  When I scale down the number of coordinators or db servers, they remain 
as part of the cluster, marked as "shutdown".  How do I permanently remove 
a member from the cluster?  Assuming that can be done, does ArangoDB then 
automatically shuffle the data around (assuming replication factor >=2) to 
other cluster nodes?
2.  What happens when say, a DB node goes down, then comes back up several 
minutes later?  Does it "catch-up" to the changes that have occurred 
elsewhere in the cluster while it was dead?  Is it helpful or harmful that 
it comes back with the data it had when it crashed?  What I mean by this 
is, will the node simply discard all of the old data anyway, then rejoin 
the cluster as a new node, and then let the rebalancing happen?  Or does 
picking up where it left off and "catching-up" help things, since there is 
less data to move around?  What about the data that was presumably 
rebalanced to other nodes when this node went down?

Any insights are greatly appreciated.

-- 
You received this message because you are subscribed to the Google Groups 
"ArangoDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to